小麦赤霉病近地遥感诊断方法及建模研究

Wheat scab is one of the most important diseases in wheat production globally, poses a serious threat to China's grain production and quality. Due to the wheat ear infection, the conventional remotely sensed data of aviation/satellite cannot effectively monitor the disease. How to identify and diagnose the Fusarium head blight (FHB) accurately has become an urgent problem to be solved in the current wheat production. This project introduces the multi-angle observation with the imaging technique of “combining spectra and imagery as one” to explore a new remotely sensed means to realize the aim. The main contents of this study includes: 1) to clear the optical diagnosis mechanism of the FHB through analyzing feature changes of spectra and images that collected from the key growth stages. And then spectral and imagery characteristics of sensitive to disease were mined and collaborative analyzed, characteristic indexes of disease identification were screened out. 2) to analyze the effect factors i.e. observation angle, illumination difference, segmentation of canopy leaves and spikes which can lower the accuracy of disease discrimination by multi-angle observation. 3) on the basis of above studies, a universal diagnosis model was established and verified through comprehensive characteristics of spectra and images with optimal observation angles. The outcomes of the project would provide technical support for accurate identification and diagnosis of the disease, reducing pesticide inputs, and risk reduction of pathogens producing toxins threat to human and animal health and so on. Meanwhile, the corresponding methods and models may be referenced for further studies on remotely sensed monitoring of spike diseases on other crops.

小麦赤霉病是全球小麦生产中最重要的病害之一，对我国粮食产量和品质构成严重威胁。由于其危害穗部，常规的航空/卫星遥感无法对该病进行有效监测。如何实现该病准确识别与诊断，成为当前小麦生产中迫切需要解决的问题。本项目借助“图谱合一”的光谱成像技术，引入多角度观测方式，探索一种近地遥感诊断穗部病害的新手段，拟开展下述研究：1)通过解析病害发生发展过程图谱特征变化，明晰穗部病害光学诊断机理；对病害敏感的图谱特征挖掘与协同分析，筛选出识别病害的图谱特征指标；2)结合多角度观测针对性分析观测角度、光照差异、冠层叶片和麦穗分割效果等因素对病害识别的影响；3)基于上述研究，综合有效观测角度信息与优选的图谱特征指标，建立具有较强普适性的病害诊断模型并验证。项目研究结果可为该病准确识别与诊断、减少农药投入、降低病菌产生毒素威胁人类和动物健康的风险等提供技术支持；同时，也为其他作物穗部病害遥感监测研究提供参考。

小麦是安徽省第一大粮食作物，常年播种面积约4300万亩，居全国第3位。赤霉病是制约安徽省乃至全国小麦提质增效的头号病害，近10年来，该病害在我省出现过4次严重流行和2次中度流行。赤霉病的发生不仅会造成小麦严重减产，更会恶化籽粒品质，降低种用价值。赤霉病危害小麦后，可产生多种毒素，严重威胁人畜健康和生命安全。由于缺乏丰产抗病的小麦品种，且江淮地区在小麦扬花灌浆期间出现连续阴雨天气的频率高，病菌群体对多菌灵等传统药剂已产生抗药性，给赤霉病化学药剂高效防控增加了难度，导致打药次数增加、用药量加大、防效降低、农药残留风险升高，且小麦产品中的毒素污染水平和膳食摄入暴露风险逐年加大，极大影响我省小麦的产量和质量安全。.本项目以安徽小麦优质生产为目标，利用空-天-地一体化的传感器技术、大数据技术、人工智能等采集并建立小麦病害、品质等指标的图谱特征库，开展数据挖掘与分析工作，构建数据驱动的决策支持模型，从而促进安徽省粮食作物生产数字化水平的提升，也为我国的粮食丰收和食品安全提供技术和方法参考。.本项目经过四年的科学研究，构建了适用于小麦单穗、群体的快速无损病情监测算法模型3个；建立了染病籽粒无损检测模型2个；开发出田间病情调查的便携式、固定节点式智能化数据采集装置3个，搭建了赤霉病大数据综合分析平台，实现了病情实时监测的信息化与智能化，为小麦赤霉病防控精准用药、药效评估、提升抗病品种选育效率提供技术支撑。上述方法与系统的研发填补了如今市场关于这方面的欠缺，可为农作物病害防治装置的开发提供技术参考，而且具有相当的盈利空间，具有广阔的市场前景。